|
BobHoward posted:SHAME
|
# ? Jul 12, 2017 23:24 |
|
|
# ? Apr 25, 2024 20:09 |
|
lol if you dont just buy aoyue crap
|
# ? Jul 12, 2017 23:47 |
|
BobHoward posted:SHAME same except from the local surplus electronics store
|
# ? Jul 12, 2017 23:50 |
|
the hakko 936 is the thinking man's weller 51. an elegant soldering station for a more civilized age
|
# ? Jul 12, 2017 23:52 |
|
just lol if you can't play tetris on your soldering iron https://www.youtube.com/watch?v=Buzew1z1AhQ get with the times you gramposaurs
|
# ? Jul 12, 2017 23:57 |
|
Sagebrush posted:the hakko 936 is the thinking man's weller 51. an elegant soldering station for a more civilized age playskool babby's first iron
|
# ? Jul 13, 2017 00:02 |
|
Raluek posted:playskool babby's first iron no thats the hakko 888
|
# ? Jul 13, 2017 00:04 |
|
the 936 is sleek and black and rectilinear, like a bauhaus fountain pen
|
# ? Jul 13, 2017 00:15 |
|
Olivil posted:no thats the hakko 888 I love my soldering iron that looks like a rejected SGI design from the 90s
|
# ? Jul 13, 2017 00:21 |
|
Sagebrush posted:the 936 is sleek and black and rectilinear, like a bauhaus fountain pen the metcal mx500 is a black monolith sent by the gods to fill us with awe
|
# ? Jul 13, 2017 09:14 |
|
I need to put down the video games for a while and pick up an idiot spare time project. any suggestions? I do not presently have a soldering station tho
|
# ? Jul 13, 2017 13:48 |
|
xoxbox?
|
# ? Jul 13, 2017 15:59 |
|
hm don't have any midi controllers but I should look into getting one for my y2k pc
|
# ? Jul 13, 2017 16:34 |
|
write a Macintosh program like me
|
# ? Jul 13, 2017 16:35 |
|
Silver Alicorn posted:I need to put down the video games for a while and pick up an idiot spare time project. any suggestions? I do not presently have a soldering station tho take up plein air oil painting, op
|
# ? Jul 13, 2017 16:49 |
|
aww yes, my yospos now has a background color gradient still messin around code:
|
# ? Jul 13, 2017 16:58 |
|
playing around with some neopixel strips, i worked out how to drive them using a hardware SPI controller instead of the bit banging solution most libraries use. to recap, the strips use a self clocking signal where the duty cycle defines 1/0, around 30-45% is 0 and 55-70% or so is 1 not clocking the bus for about 9 µs or more causes a bus reset that latches the data. the bus speed is not extremely tight but must be stable (approx 800 kHz) using some strategic time delay elements it's possible to take a standard SPI signal, generate a 90 degree shifted clock, a stretched in-phase clock, and a slightly time delayed data signal. a quad nand gate then implements some combinational logic that generates the final output signal basically the 0 state is a short pulse, generated from the combination of clock AND NOT quadrature clock the 1 state is the 0 state + some more on-time, so the 1 state output is the delayed data AND the time streched in-phase clock those two are logically ORed to generate the final output entire solution requires 6 inverters and 4 NAND gates, along with 4 R-C filters real hardware logic trace tested with an STM32F103: top is the output; the 0 state is SCK AND NOT SCK_Q the 1 pulse is the 0-state ORed with (D_D (data delayed) AND S_D (SCK_delayed, the stretched clock). the data delay in this shot is a little much, but just small enough that no glitches show up the big reason to do this is because it's a fun challenge but also because the bit-banging requires interrupts to be disabled a lot on the MCU. the MCUs I'm targeting are guaranteed to have hardware SPI with DMA so with this setup the entire strip can be updated with no extra RAM usage and without interfering with other peripherals. with an extra gate or two it could even have a chip select, but most MCUs have a spare SPI so not really necessary. the strips will not run if the duty cycle of a 1-state is 50%, if they could accept 25/50% as 1/0 then the pulse stretcher and data delay could be removed.
|
# ? Jul 13, 2017 17:49 |
|
bigclivedotcom is that you? nah but seriously, awesome stuff!@
|
# ? Jul 13, 2017 18:26 |
|
obviously you could just encode the data as triplets instead of bits and DMA that at 3x the speed (which someone else already wrote code to do) but that requires at least 3x the ram, and probably 4x if you need a frame buffer to work on, and the encoding process takes some time too. for very large strips that can add up, for my 300 led strip that's 900 bytes for a full RGB frame buffer vs. 3600 bytes for an RGB buffer + the DMA buffer. running it on a STM32F030 with 4k of RAM that makes a big difference!
|
# ? Jul 13, 2017 20:20 |
|
i like how you worked out having different duty cycles. everyone seems to disregard that when messing with waves
|
# ? Jul 13, 2017 20:37 |
|
you can buy SPI led strips (APA102) that are functionally identical to the neopixels (WS2812) in every way except the protocol https://www.adafruit.com/product/2241 but that is still a v cool bit of reverse engineering
|
# ? Jul 13, 2017 21:41 |
|
i had a bit of fun with them but had to use a lookup table because their response was so far from linear it wasn't even funny.. ugh and dim colours were hosed. bright colours were fantastic but I always find with poo poo like that there seems to be orders of magnitude more difference in brightness between 0-1 and between like 100-255 or whatever. I dont see in theory why you cant drive them with such a low pwm that you can do a proper fade to black rather than to dim and then off with nothing inbtween. really hosed up my plans
|
# ? Jul 14, 2017 01:37 |
|
you can get pretty drat close to a linear fade-out if you just use a proper gamma table, which it sounds like you did, but yeah you're always gonna be limited by the resolution since most of these things only take 8 bits per color channel. also the context matters. you can get a pretty good looking fade to nothing if you have the LEDs running in a regular room with the lights on, and you're comparing it to the ambient level. but in the darkness, where only the LEDs are showing, your eyes are super sensitive. there's research that suggests that people can detect a single photon when they're properly accommodated for the dark. that level of control is asking too much of a 10 cent LED imo
|
# ? Jul 14, 2017 02:20 |
|
some interesting facts no doubt how hard would it be to increase the resolution? is there a practical limit to how fast you can PWM?
|
# ? Jul 14, 2017 03:11 |
|
if it's an rgb-on-a-chip like the WS2812 or APA102, you can't change the pwm frequency -- it's all done onboard. you send 24 bits of color data and the LED does the rest. so even if the frequency technically allows you to have better than 1/256 resolution, you can't change it. iirc the pwm frequency on the neopixels is around 100khz, which at 8 bits of resolution means a 50% duty cycle period would be like ~400hz. if you're doing it yourself with discrete diodes, then the hard limit i guess would be the minimum transition time for the LED itself, which is gonna be nanoseconds. pwm of a few megahertz wouldn't be hard to achieve (note that this would only work for colored diodes -- white ones use a blue LED to excite a white phosphor, and while the phosphor turns on instantly, it has a noticeable fade-out time, like an incandescent bulb) Sagebrush fucked around with this message at 03:24 on Jul 14, 2017 |
# ? Jul 14, 2017 03:22 |
|
the phosphor effect would be beneficial no?? cool
|
# ? Jul 14, 2017 04:09 |
|
Sagebrush posted:you can buy SPI led strips (APA102) that are functionally identical to the neopixels (WS2812) in every way except the protocol neopixels have some really big advantages though, namely a) cheap and b) non-expensiveness. my 5m roll was like $25 with 60 leds/m echinopsis posted:some interesting facts no doubt the fastled library includes temporal dithering by default, i tried it and it definitely helped but there are limits. it could work well for short strings (or say a bunch of shorter pieces in parallel) since the string write-time goes down and more dither-updates can be squeezed in per second, giving more resolution. it also has nice colour modes and can apply a colour correction to all your LEDs which is pretty nice, helps keep them approximately neutral white instead of completely blue-white. with correction on it does generate a bit of colour casting through the brightness range, especially for very low intensities. a better dither could help in theory but if that's a major issue then really you just need a higher resolution chip.
|
# ? Jul 14, 2017 06:14 |
|
Silver Alicorn posted:I need to put down the video games for a while and pick up an idiot spare time project. any suggestions? I do not presently have a soldering station tho You could build yourself a soldering station.
|
# ? Jul 14, 2017 09:05 |
|
LP0 ON FIRE posted:aww yes, my yospos now has a background color gradient user scripts own code:
|
# ? Jul 14, 2017 11:12 |
|
Wheany posted:user scripts own noice LP0 ON FIRE fucked around with this message at 14:22 on Jul 14, 2017 |
# ? Jul 14, 2017 14:15 |
|
the most idiot project of all time award goes to https://soundcloud.com/danwarren/yakety-sad
|
# ? Jul 15, 2017 08:47 |
|
not all that impressive, but nifty looking. made a little visual debugger for a quadtree game level splitter/merger. the idea is it will take the tiled game level, merge identical areas into chunks that are efficient for visibility testing & rendering. Doc Block fucked around with this message at 05:02 on Jul 23, 2017 |
# ? Jul 23, 2017 04:47 |
|
guess I'll get back into classic Atari programming... where did I leave my mac/65 disks...
|
# ? Jul 24, 2017 01:56 |
|
is there an OS that support 1k+ controller inputs across 1 or more hid devices
|
# ? Jul 24, 2017 02:13 |
|
moron izzard posted:is there an OS that support 1k+ controller inputs across 1 or more hid devices like >1000 keyboard buttons?
|
# ? Jul 24, 2017 02:23 |
|
any digital input really. 1000 unique inputs you could pick up in like a Unity engine game. There was someone who made a 100 button controller for an install this week at the Gallery Hop and mentioned in passing that they wanted to scale it up.
moron izzard fucked around with this message at 02:26 on Jul 24, 2017 |
# ? Jul 24, 2017 02:24 |
|
moron izzard posted:is there an OS that support 1k+ controller inputs across 1 or more hid devices windows directinput supports 128 buttons per controller, and you can have up to 127 devices per usb bus, so that's 16256 buttons i guess?
|
# ? Jul 24, 2017 03:02 |
|
Silver Alicorn posted:guess I'll get back into classic Atari programming... where did I leave my mac/65 disks... my 6502 emulator is almost done then I can move on to adding a debugger and adding Apple 1 hardware
|
# ? Jul 24, 2017 03:21 |
|
moron izzard posted:is there an OS that support 1k+ controller inputs across 1 or more hid devices i am pretty sure this is literally what midi was designed for
|
# ? Jul 24, 2017 03:21 |
|
|
# ? Apr 25, 2024 20:09 |
|
Corla Plankun posted:i am pretty sure this is literally what midi was designed for yeah, and if not midi, osc
|
# ? Jul 24, 2017 03:23 |